A simple critical state based double-yield-surface model

for clay behavior under complex loading

Zhenyu Yin, Qiang Xu, Pierre-Yves Hicher

To cite this version:

Zhenyu Yin, Qiang Xu, Pierre-Yves Hicher. A simple critical state based double-yield-surface model
for clay behavior under complex loading. Acta Geotechnica, 2013, 8 (5), pp.509-523. �10.1007/s11440-
013-0206-y�. �hal-01004941�

HAL Id: hal-01004941

https://hal.science/hal-01004941
Submitted on 8 Apr 2017

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
Public Domain
A simple critical-state-based double-yield-surface model for clay
behavior under complex loading
Zhen-Yu Yin • Qiang Xu • Pierre-Yves Hicher

Z.-Y. Yin
Department of Civil Engineering,
Shanghai Jiao Tong University,
Shanghai 200240, People’s Republic of China

Z.-Y. Yin  Q. Xu
State Key Laboratory of Geohazard Prevention
and Geoenvironment Protection, Chengdu University
of Technology, Chengdu 610059, People’s Republic of China

Z.-Y. Yin  P.-Y. Hicher

Research Institute in Civil and Mechanical Engineering,
UMR CNRS 6183, Ecole Centrale de Nantes,
44300 Nantes, France

Abstract The critical state concept has been widely used 1 Introduction
in soil mechanics. The purpose of this study is to apply this
concept in the framework of multi-mechanism elastoplas- The mechanical behavior of clay has been widely investi-
ticity. The developed model has two yield surfaces: one for gated by numerous researchers during the last decades [3,
shear sliding and one for compression. In this model, the 8, 11, 18, 20, 23, 33, 42]. A clayey material is contractive
location of the critical state line is explicitly considered and (decreasing void ratio in drained condition in Fig. 1a and
related to the actual material density to control the peak decreasing mean effective stress in undrained condition in
strength and the phase transformation characteristics. The Fig. 1b) in a normally or lightly over-consolidated state
stress reversal technique is incorporated into the model for and becomes dilative (increasing void ratio in drained
describing clay behavior under complex loading including condition in Fig. 1a and increasing mean effective stress in
changes of stress direction. The determination of the model undrained condition in Fig. 1b) when the over-consolida-
parameters is discussed; it requires only one drained or tion ratio increases, with the stress state and the void ratio
undrained triaxial test up to failure with an initial isotropic converging to the critical state (CSL in p0 –q plane and CSL
consolidation stage. The model is used to simulate drained in e–logp0 / plane, shown in Fig. 1a, b). The location of
and undrained tests under monotonic loading with different the CSL in the e–logp0 / plane is a key factor for deter-
over-consolidation ratios on various remolded and natural mining the magnitude of contraction or dilation (see dif-
clays, including true triaxial tests with different Lode’s ference between B and B0 , C and C0 in Fig. 1c) from an
angles. Drained and undrained tests under cyclic loadings initial given void ratio.
are also simulated by using the set of parameters deter- Based on experimental observations, conventional el-
mined from monotonic tests. The comparison between astoplastic models for clay have been developed, and
experimental results and numerical simulations demon- generally can be classified into two categories:
strate a good predictive ability of this new simple model. (1) Critical-state-based elastoplastic models [4, 9, 20, 21,
23, 25–27, 30, 33, 34, 36, 41]. In these models, the
slope of CSL in the e–logp0 / plane ( k ) w is
explicitly used as a parameter, and the location of this
Keywords Clay  Constitutive modeling  Critical state  CSL is implicitly governed by the intersection point
Cyclic loading  Elastoplasticity of the yield surface with the CSL in p0 –q plane.

1
Fig. 1 Schematic plot for the mechanical behavior of clay a drained behavior; b undrained behavior; c influence of critical state line

(2) Multi-mechanism models [13, 14, 16, 19, 24]. The reversal technique into the model is then presented in order
model by Hujeux [16] used the location of CSL as to be able to simulate tests with shear stress reversal.
presented in the first category, and twelve parameters Finally, the model is validated by comparing experimental
are needed. In other models, the CSL in the results and numerical simulations of drained and undrained
e–logp0 / plane has not been used. As a result, the tests on isotropically and anisotropically consolidated clay
contraction and dilation up to critical states during samples with different OCRs (over-consolidation ratio)
compression and shearing are not guaranteed. Thus, under both monotonic and cyclic loadings.
these models are difficult to simulate clay behavior
along various stress paths and loading conditions.
2 Constitutive model
Different from them, other models have also been
developed for clay. For instance, Mašin [22] and Huang According to elastoplasticity theory, the total strain rate is
et al. [15] proposed clay models based on the hypoplas- additively composed of the elastic strain rates and plastic
ticity framework; Calladine [5] and Schweiger et al. [28] strain rates
proposed clay models based on the concept of multi-lam-
e_ ij ¼ e_ eij þ e_ pij ð1Þ
inate modeling. In these models, the CSL in the e–logp0 /
plane has not been used, thus the contraction and dilation where e_ ij denotes the (i,j) component of the total strain rate
up to critical states cannot be guaranteed. tensor, and the superscripts e and p stand, respectively, for
In sand modeling, the critical state concept has been the elastic and plastic components.
explicitly used [2, 10, 31]. This explicit method has also
been adopted by the authors to account for the interlocking 2.1 Elastic behavior
effect and neighboring effect of sand grains and clay
aggregates at inter-particle contact level in different ver- The elastic behavior is assumed to be isotropic, as follows:
sions of the micromechanics-based model [6, 7, 37–40].
1þt 0 t
Good predictive performance has been achieved for both e_ eij ¼ rij  r0kk dij ð2Þ
E E
sand and clay, and note that the models are physically
attractive on describing the behavior of anisotropy. How- where t and E are Poisson’s ratio and Young’s modulus; r0ij
ever, the micromechanics-based models are usually com- is the effective stress tensor; dij is Kronecker’s delta.
putational demand. Therefore, this paper is an attempt to E can also be replaced by the elastic bulk modulus K by
explicitly use the CSL in the double-yield-surface-model- E ¼ 3Kð1  2tÞ. For clay, the elastic bulk modulus K can
ing method in order to propose a model simple to use and be determined by the relationK ¼ p0 ð1 þ e0 Þ=j, with the

at the same time well adapted to the modeling of clay mean effective stress p0 ¼ r0kk 3 (see Roscoe and Burland
behavior. [26]), the slope of the swelling line j and the initial void
In the following sections, the development of the dou- ratio e0. Assuming a common value of t = 0.25 for clay,
ble-yield-surface model using the critical state concept for only the parameter j is then needed for the elastic
clay is firstly presented. The implementation of the stress behavior.

2
2.2 Plastic behavior the value of pcr0 is taken equal to 100 kPa. The critical state
line can then be defined by the two parameters ecr0 and k.
Based on conventional elastoplasticity theory, the plastic Using the critical state concept, the density state of the
strain can be obtained, as follows: soil is defined as the ratio ec/e, where e is the actual void
og ratio and ec is the critical void ratio obtained by Eq. (5) for
e_ pij ¼ dk ð3Þ the value of p0 at the given stress state (see Fig. 2 for the
or0ij
state A). This implies: ec/e \ 1 for normally consolidated
where dk is plastic multiplier depending on the stress rate and slightly over-consolidated clay, ec/e [ 1 for heavily
and the plastic hardening law; g is plastic potential. over-consolidated clay, and ec/e = 1 when the initial state
The proposed model uses two yield surfaces: one for of clay lies on the CSL in the e–logp0 plane.
shear sliding (fS) and one for compression (fC) (see Fig. 2).
Equation (3) can be expressed as follows: 2.2.2 Shear-sliding criterion
ogS ogC
e_ pij ¼ e_ pS pC
ij þ e_ ij ¼ dk
S
0 þ dkC 0 ð4Þ In several models for sand and silt [10, 17, 32], the shape of
orij orij
the yield surface for the plastic shear component is linear in
where the superscripts S and C represent the shear sliding the p0 –q plane. Here, we adopt the same shear criterion for
and compression components, respectively. For one com- clay, rwritten
ffiffiffiffiffiffiffiffiffiffiffiffi as follows:
ponent with f \ 0, the corresponding dk is taken equal to 3
fS ¼ rij rij  H ð6Þ
zero. 2
where rij ¼ sij =p0 with sij ¼ r0ij  p0 dij ; H is the hardening
2.2.1 CSL-related density state
parameter defined by a hyperbolic function in the H  epd
One of the important elements to be considered in soil plane given by
modeling is the critical state concept. At critical state, the Mp epd
H¼  ð7Þ
clay material remains at a constant volume while it is 1 Gp þ epd
subjected to a continuous distortion. The void ratio corre-
sponding to this state is ec. The critical void ratio ec is a where Gp controls the initial slope of the hyperbolic curve
function of the mean effective stress p0 . The relationship g  epd ðg ¼ q=p0 Þ. Equations (6) and (7) guarantee that the
has traditionally been written as follows: stress ratio g reaches the peak stress ratio Mp.
 0  According to Biarez and Hicher [3], the peak friction
p  
ec ¼ ecr0  k ln ð5Þ angle /p (related to Mpc ¼ 6 sin /p 3  sin /p ) depends
pcr0
on the internal friction angle /l (related to the critical state
The critical state line in the e–logp0 plane is explicitly  
value Mc ¼ 6 sin /l 3  sin /l ) and the density state of
located by three parameters: ecr0 and pcr0 determine a ref- soil (ec/e).
erence critical state point in the e–logp0 plane and k
ec
determines the slope of critical state line. For convenience, tan /p ¼ tan /l ð8Þ
e

3 Equation (8) shows that, in a loose structure, the peak

q= sij sij CSL
2 friction angle /p is smaller than /l . On the other hand, a
M
dense structure provides higher degree of interlocking.
A
Thus, the peak friction angle /p is greater than /l . When
fC the loading stress reaches the peak friction angle /p , the
fS
dense structure dilates and the degree of interlocking
Elastic region p’ relaxes. As a consequence, the peak friction angle is
log(p’)
reduced, which results in a strain-softening phenomenon.
A Adopting the idea of the OCR effect on the shear
e Undrained path
modulus proposed by Hardin [12], Gp is considered as a
Drained path function of OCR as follows:
ec
Gp ¼ Gp0 OCR2 ð9Þ
CSL
e
where Gp0 is an input parameter instead of Gp; OCR can be
Fig. 2 Principle of the critical state double-yield-surface model for computed based on the size of the compression yield sur-
clay face as shown in the next section.

3
 In order to take into account dilation or contraction J2 and J3 are the second and the third invariants of the

during shear sliding, a non-associated flow rule is intro- deviatoric stress tensor, given by J2 ¼ sij sij 2;J3 ¼

duced. We propose an explicit derivation of the potential sij sjk ski 3.
surface, given by
og og op0 og osij 2.2.3 Compression criterion
¼ 0 0 þ with
orij op orij osij or0ij
0
rffiffiffiffiffiffiffiffiffiffiffiffi! rffiffiffi In order to describe the compressible behavior of clay, a
og 3 og 3 rij second yield surface is added. The second yield function is
¼ D Mpt  rij rij ; ¼ pffiffiffiffiffiffiffiffi ð10Þ
op0 2 osij 2 rij rij assumed to be as follows:
which implies the following stress-dilatancy relationship 3 sij sij
fC ¼ p02 þ  p2c ð14Þ
depv   2 R2
p ¼ D Mpt  g ð11Þ
ded where the constant R controls the shape of the yield surface
and pc is the hardening parameter controlling the size of the
where the contractancy or dilatancy depends on the sign of yield surface. The yield surface expands with the plastic
‘‘Mpt - g’’; and D is a material constant controlling the volumetric strain. The hardening rule of the Cam-Clay
magnitude and evolution of the contractancy or dilatancy. model is adopted:
Mpt is the slope of the phase transformation line for clay
which can be derived from the friction angle /l by 1 þ e0 p
dpc ¼ pc de ð15Þ
assuming a formula similar to the peak friction angle kj v

e 1 An associated flow rule is adopted for the compression

c behavior.
tan /pt ¼ tan /l ð12Þ
e Parameter R can be derived as follows:
Equation (10) combined with Eq. (12) implies that in a One-dimensional straining corresponds to
loose structure with e [ ec, the phase transformation  
  ded 2
angle /pt (related to Mptc ¼ 6 sin /pt 3  sin /pt ) is ¼ ð16Þ
dev K0 3
bigger than /l ; in a dense structure with e \ ec, the phase
transformation angle /pt is smaller than /l which allows Assuming that elastic strains are much smaller than plastic
strains, Eq. (13) can be approximated by
the dense structure to be first contractive and then dilative
 p
during deviatoric loading. For both loose and dense ded 2
¼ ð17Þ
structures, when the stress state reaches the critical state depv K0 3
line, the void ratio e becomes equal to the critical void
ratio ec, then zero dilation or contraction takes place. Combining Eq. (14) with the flow rule derived from Eq.
Thus, the constitutive equations guarantee that stresses (11), for the condition of one-dimensional consolidation, R
and void ratio reach simultaneously the critical state in can be expressed by
the p0 –q–e space. rffiffiffiffiffiffiffiffiffiffi
3
In order to interpolate Mp and Mpt between their values R¼ g ð18Þ
2 K0
for compression and those for extension by means of the
Lode’s angle h (see Sheng et al. [29]), the following where gK0 ¼ 3Mc =ð6  Mc Þ for normally consolidated clay
 
expressions of Mp and Mpt are proposed as follows: adopting Jacky’s formula K0 ¼ 1  sin /l . Therefore, R
" #14 is not needed for input.
2c41
Mp ¼ Mpc   ;
1 þ c41 þ 1  c41 sin 3h 2.3 Incorporation of shear stress reversal technique
" #14
2c42 The above constitutive equations were developed based on
Mpt ¼ Mptc   ð13Þ
1 þ c42 þ 1  c42 sin 3h experimental clay behavior, considering isotropically con-
     solidated samples under monotonic loading. When the
where c1 ¼ 3  sin/p 3 þ sin/p and c2 ¼ 3  sin/pt
  direction of the shear stress changes during the loading, the
3 þ sin/pt , assuming the same peak friction angle and stress reversal effect on the stress–strain relationship needs
phase transformation angle for different Lode’s angle; the to be incorporated.
 pffiffi 
Lode’s angle is expressed as p h ¼ 1
sin 1 3 3J3
 p6. Let us denote the stress state and plastic strain state at
6 3 3=2
2J2 the moment of shear reversal to be the residual stresses

4
 . .
(marked as r0R ij ) and the residual plastic strains (marked as where rijR ¼ r0R p0R  dij (with p0R ¼ r0R 3) is the mobi-
ij ij
R
eij ). These two mechanical state variables have a significant lized stress ratio at the moment of stress reversal; epR ij ¼
influence on the subsequent shear behavior (see Fig. 3). As pR pR 
eij  ekk dij 3 is the plastic deviatoric strain tensor at the
a consequence, the yield function (see Eq. (6)), the hard-
moment of stress reversal. The estimation of Lode’s angle
ening rule (see Eq. (7)) and the flow rule (see Eq. (10))  pffiffi 
3 3J30
have been revised as follows: to calculate Mp and Mpt is modified as h ¼ 13 sin1 03=2
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi . .
2J2
3 
fS ¼ rij  rijR rij  rijR  H ð19Þ with J20 ¼ rij0 rij0 2; J30 ¼ rij0 rjk0 rki0 3 where rij0 ¼ rij  rijR .
2
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
Equation (20) indicates that the same form of hardening
Mp ep
d p 2 p pR

p pR rule can be used for both loading and unloading, but that
H¼ with ed ¼ e  eij eij  eij
Gp þ ep d 3 ij some scaling process on the values of Mp is required.
ð20Þ Eq. (21) implies that the amount of dilation is different
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ! upon shear reversal. This concept is similar to that pro-
og 3  
¼ D Mpt 
 rij  rijR rij  rijR ; posed by Balendran and Nemat-Nasser [1] and Gajo and
op0 2 Muir Wood [10]. This stress reversal technique is similar to
 that used in Masing’s rule, in bounding surface plasticity
rffiffiffi
og 3 rij  rijR
¼ nij with nij ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð21Þ (Taiebat and Dafalias [31]) and in subloading surface
osij 2  
plasticity (Yamakawa et al. [35]).
rij  rijR rij  rijR
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2.4 Determination of model parameters
3 
Mp ¼ Mp nij  rijR Mp nij  rijR ;
r2ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð22Þ The model contains six material parameters divided into
 3 
four groups (see Table 1). Based on their physical mean-
Mpt ¼ Mpt nij  rijR Mpt nij  rijR
2 ings, these parameters can be determined as follows:
(1) the slope of the swelling line j can be measured from
the unloading curve of an isotropic compression test;
(2) the internal friction angle /l can be obtained from Mc
Reverse point measured from drained or undrained triaxial tests up
 
R to failure /l ¼ arcsin½3Mc =ð6 þ Mc Þ ;
3 R R
R rij rij (3) the slope of the CSL k is assumed to be the same as
2
the slope of the isotropic compression line and can,
pR p
d d therefore, be measured from an isotropic compression
test;
* 3
rij rijR rij rijR
(4) the position of the critical state in the e–logp0 plane
2 pR pR 2
d
pR
eij eij can be obtained from drained or undrained triaxial
3
tests up to failure; then, the reference critical void
Current point
ratio ecr0 corresponding to p0 = 100 kPa can be
p*
d
2 p
eij eijpR eijp eijpR measured in the e–logp0 plane;
3
(5) the plastic stiffness Gp0 can be obtained by curve
Fig. 3 Principle of stress reversal in loading and unloading fitting from the deviatoric stress–strain curve at small
strain level (see Fig. 4a for the effect of Gp0 on the

Table 1 Values of model parameters for simulated clays

Group Parameter Definition Black White Fujinomori Mixture BBC LCT

Elasticity j Slope of swelling line 0.079 0.034 0.02 0.034 0.023 0.008
Critical state k Slope of compression line 0.244 0.089 0.093 0.173 0.171 0.066
ecr0 Reference critical void ratio 1.33 0.653 0.77 0.977 0.986 0.435
/c Critical friction angle 21.4° 23° 34° 19.2° 35° 30°
Plastic stiffness Gp0 Plastic hardening modulus 50 70 115 40 155 170
Dilatancy D Dilatancy constant 1 1 1.2 0.3 1 0.8

5
 900 900 0
(a) (b) (c)
1
600 600
q (kPa)

q (kPa)
2

(%)
Gp0 = 0.005 D = 0.5
D = 1.0
D = 1.0

v
D = 0.5 3
300 300
Gp0 = 0.015
D = 1.5 4
Gp0 = 0.045 D = 1.5

0 0 5
0 1 2 3 4 0 300 600 900 0 10 20 30
1 (%) p' (kPa) 1 (%)

Fig. 4 Parametric study a Gp0 effect on the deviatoric stress versus axial strain for undrained triaxial test on NC clay; b D effect on the effective
stress path for undrained triaxial test on NC clay; c D effect on the volumetric strain versus axial strain for drained triaxial test on NC clay

deviatoric stress–strain curve, keeping other param- 150 CL: Low plastic inorganic clays, sandy and silty clays
Black clay
eters (White clay in Table 1) constant); OL: Low plastic inorganic or organic silty clays
CH: High plastic inorganic clays White clay
(6) the dilatancy constant D can be finally obtained by 100 OH: High plastic fine sandy and silty clays
Fujinomori clay
curve fitting from the effective stress path of an

Ip
U-line: I p = 0.9(wL-8) CH Clay mixture
undrained test (see Fig. 4b for the effect of D on the 50
CL Boston Blue clay
effective stress path, keeping other parameters (White
OH A-line: I p = 0.73(wL-20) Lower Cromer Till
clay in Table 1) constant) or, alternatively, from the 0
OL
evolution of the volumetric strain during a drained 0 50
wL
100 150

test (see Fig. 4c for the effect of D on the evolution of

the volumetric strain). Fig. 5 Classification of clays by liquid limit wL and plasticity index
Ip
Additionally, the model involves two state variables (the
void ratio e and the size of the compression yield surface
pc) which require their initial values to be determined. e0 Drained tests on Fujinomori clay and undrained tests on
can be measured from the tested sample. The initial size pc0 clay mixture under cyclic loadings were also simulated for
is determined by the consolidation history of the tested the validation.
sample (i.e., the consolidated stress for reconstituted clay
and the preconsolidation stress for natural clay). 3.1 Different over-consolidation ratios
In summary, all values of the model parameters and
state variables can be easily determined based on one 3.1.1 Black clay
drained or undrained triaxial test up to failure with an
isotropic consolidation stage. Drained triaxial tests on black kaolinite clay samples were
performed by Zervoyanis [42]. Four tests began with an
isotropic consolidation up to 800 kPa, then three of them
3 Experimental verification were unloaded to 400, 200, and 100 kPa, respectively
(OCR = 1, 2, 4, and 8), then followed by axial loading up to
The experimental verification is presented herein with failure under drained condition. The tested Black clay is a
reference to drained and undrained tests results under remolded clay prepared from a slurry obtained by mixing
monotonic loading on various remolded and natural clays clay powder and water at a water content equal to two times
(Black clay by Zervoyanis [42]; White clay by Biarez and the liquid limit. The parameters presented in Table 1 were
Hicher [3]; Fujinomori clay by Nakai and Hinokio [23]; calibrated from one triaxial compression test on the normally
Boston blue clay by Ladd and Varallyay [18]; Lower consolidated specimen with an isotropic consolidation stage.
Cromer Till by Gens [11]; and clay mixture by Li and In Fig. 6a, the stress–strain curves for OCR = 1 and 2
Meissner [20]). These tests were performed on clays of show a continuous strain hardening due to a continuous
different mineral contents and Atterberg limits. Figure 5 contractancy corresponding to the paths in Fig. 6c above
shows the classification of these clays using Casagrande’s the critical state line, and the stress–strain curves for
plasticity chart. According to this chart, the selected OCR = 4 and 8 show strain hardening followed by strain
experimental results consist of both low and high plastic softening corresponding to the paths in Fig. 6c below the
inorganic clays, as indicated in Fig. 5. critical state line, as respected by using Eq. (8). The pre-
dicted void ratio changes in Fig. 6b, c show a contractive

6
 1000 1.4 1.4
OCR=1 OCR=1 OCR=2 IC test
OCR=2 OCR=4 OCR=8 CSL
OCR=4 Model
750 1.2 1.2 OCR=1
OCR=8 OCR=2
Model
q (kPa)
OCR=4

e
OCR=8

e
500 1 1
Model

250 0.8 0.8

(a) (b) (c)

0 0.6 0.6
0 10 20 30 0 10 20 30 100 1000 10000
1 (%) 1 (%) p' (kPa)

Fig. 6 Comparison between experimental results and model predictions for drained triaxial tests on Black clay with different OCRs: a deviatoric
stress versus axial strain; b void ratio versus axial strain; c void ratio versus mean effective stress

behavior for OCR = 1 and 2, and a dilative behavior for over-consolidated White clay were reported by Biarez and
OCR = 4 and 8, as respected by using Eq. (12). The paths Hicher [3]. In the case of drained tests, specimens were
in the e–logp0 plane in Fig. 6c show that the void ratio isotropically consolidated up to 400 and 800 kPa, respec-
approaches the critical state line when the stress state tively, and then loaded to failure keeping the confining
approaches the CSL in the p0 –q plane. An overall good stresses. In the case of undrained tests, three samples were
agreement is observed between experimental and numeri- isotropically consolidated up to 800 kPa, and two of them
cal results for different OCRs. were unloaded to 400 and 67 kPa (OCR = 1, 2, 12). The
parameters presented in Table 1 were calibrated from one
3.1.2 White clay drained compression test at the confining stress r03c ¼
800 kPa with an isotropic consolidation stage.
Two drained triaxial tests on normally consolidated White Using the calibrated parameters, the predicted test
clay and three undrained triaxial tests on normally and results are plotted in Fig. 7. For drained tests, the stress–

(a) (b) (c)

1200 0.6 0.8
Experiment IC stage
'3c = 800 kPa Unloading stage
Model
Shearing stage
900 0.55 0.7
CSL
1
q (kPa)

Model
'3c = 400 kPa
600 0.5 0.6
e
e

'3c = 400 kPa

300 0.45 '3c = 800 kPa 0.5 1

Experiment
Model
0 0.4 0.4
0 10 20 30 0 10 20 30 100 1000 10000
1 (%) 1 (%) p' (kPa)
(d) (e) (f)
900 900 0.8
OCR=1 OCR=1 IC test
OCR=2 OCR=2 CSL
OCR=12 OCR=12 0.7 OCR=1
600 Model 600 Model OCR=2
q (kPa)

q (kPa)

OCR=12
0.6 Model
e

300 300
0.5

0 0 0.4
0 5 10 15 0 300 600 900 10 100 1000 10000
1 (%) p' (kPa) p' (kPa)

Fig. 7 Comparison between experimental results and model predictions for triaxial tests on White clay a deviatoric stress versus axial strain for
drained tests with OCR = 1; b void ratio versus axial strain for drained tests with OCR = 1; c void ratio versus mean effective stress for drained
tests; d deviatoric stress versus axial strain for undrained tests with different OCRs; e effective stress path for undrained tests with different
OCRs; f void ratio versus mean effective stress for undrained tests

7
strain curves in Fig. 7a and the void ratio change in experimental data on BBC. Figures 9c, d and 10a, g, j show
Fig. 7b, c shows good agreement between experimental the comparisons between measured and predicted results of
and numerical curves. For undrained tests, the computed undrained triaxial tests on isotropically consolidated sam-
and measured stress–strain curves are also in good agree- ples of LCT and BBC. For both natural clays, the present
ment (Figs. 7d–f). The effective stress paths indicate that, model gives good prediction for tests on lightly and heavily
for the normally consolidated and slightly over-consoli- over-consolidated samples using the set of parameters
dated samples, the stress paths do not overpass the critical determined from the tests on normally consolidated clay
state line; whereas, for the strongly over-consolidated samples. The predictions for tests on anisotropically con-
specimen, the stress path goes above the critical state line, solidated samples will be discussed later.
at which dilation occurs, leading to an increase in the mean
effective stress, before converging toward the critical state 3.2 Influence of Lode’s angle
line. Overall, by using parameters calibrated from one
drained test, the model is capable of predicting the stress– 3.2.1 Fujinomori clay
strain behavior of undrained tests for specimens with dif-
ferent OCRs. Drained triaxial tests on Fujinomori clay were performed at
constant p0 in compression and extension on isotropically
3.1.3 Natural clays consolidated samples with different OCRs (OCR = 1, 2, 4,
8) by Nakai and Hinokio [23]. Drained true triaxial tests on
Undrained triaxial tests on isotropically consolidated normally consolidated Fujinomori clay samples were also
samples of clay mixture with different OCRs (OCR = 1.0, performed at constant p0 with different Lode’s angles (0°,
1.6, and 4.0) were performed by Li and Meissner [20]. The 15°, 30°, 45°). The parameters presented in Table 1 were
main minerals of the clay mixture are kaolinite (60 %), calibrated from one drained test in compression on a nor-
illite (5 %), and quartz. The parameters presented in mally consolidated specimen with an isotropic consolida-
Table 1 were calibrated from the undrained triaxial test tion stage.
performed on the over-consolidated specimen (OCR = 4) Comparisons between experimental and numerical
with its consolidation stage. The determined parameters results for triaxial tests in compression and extension are
were used to simulate undrained triaxial tests on samples presented in Fig. 11. One can see that the peak stress ratio
with OCR = 1 and 1.6. As shown in Fig. 8, the model of the over-consolidated clay samples and the amount of
predictions are in good agreement with the experimental dilatancy increase with increased OCR. The model was
results. able to capture the trend of the stress–strain behavior of
Numerical simulations of tests performed on samples of clay with different OCRs under both compression and
Boston blue clay (BBC) by Ladd and Varallyay [18] were extension conditions.
also undertaken, as well as tests performed on samples of The observed and predicted responses of normally
Lower Cromer Till (LCT) by Gens [11]. The database for consolidated Fujinomori clay in true triaxial tests with
both clays includes undrained triaxial tests on both iso- different Lode’s angles are shown in Fig. 12. The stress
pffiffiffiffiffiffiffi
tropically and anisotropically consolidated specimens with ratio 3J2 p0 is plotted against the principal strains (major
different OCRs. Figure 9a, b shows the computed results of strain in Fig. 12a, intermediate strain in Fig. 12b, minor
anisotropic compression and oedometer tests compared to strain in Fig. 12c, respectively). The predictions agree

(a) 1.3 (b) 450 (c) 450

IC test OCR=1 OCR=1.6 OCR=1 OCR=1.6
OCR=4 Model OCR=4 Model
Model
1.1
300 300
q (kPa)

IC line
q (kPa)

0.9
e

150 150
Critical state
0.7

0.5 0 0
10 100 1000 0 5 10 15 20 0 150 300 450 600
p' (kPa) 1 (%) p' (kPa)

Fig. 8 Comparison between experimental results and model predictions for undrained tests on clay mixture a void ratio versus mean effective
stress for isotropic compression test, b deviatoric stress versus axial strain; c effective stress path

8
 (a) (c) (e)
1.3 1.2
OCR=1 OCR=2 K0=1(Com)
AC line, K 0=0.53 0.9
OCR=4 OCR=8
K0=1(Ext) 0.9
Model
1.1 K0=0.53(Com)
IC line 0.6
0.6

q (kPa)
K0=0.53(Ext)

q (kPa)
e

0.9 Model
Critical state 0.3

0.3
0.7 0
AC test, K0=0.53 -15 -10 -5 0 5 10
Model -0.3
0.5 0
10 100 1000 0 3 6 9 12
-0.6
p' (kPa) 1 (%) 1 (%)
(b) (d) (f)
1.3 1.2 K0=1(Com) K0=0.53(Com)
OCR=1 OCR=2
0.9 K0=0.53(Ext) K0=1(Ext)
OCR=4 OCR=8
0.9 Model
Model
1.1
0.6 0.6
q (kPa)

q (kPa)
e

0.9
0.3

0.3 0
0.7
Oedometer test 0 0.3 0.6 0.9 1.2
Model -0.3
0.5 0
10 100 1000 0 0.3 0.6 0.9
-0.6
'v (kPa) p' (kPa) p' (kPa)

Fig. 9 Comparison between experimental results and model predictions for tests on Boston blue clay a void ratio versus mean effective stress for
anisotropic compression test, b void ratio versus vertical effective stress for oedometer test, c, e deviatoric stress versus axial strain for undrained
tests; d, f effective stress path for undrained tests

reasonably well with the experimental data in all cases. The Figure 9e, f shows a good agreement between the numer-
comparisons demonstrate that the proposed model can take ical and the experimental results for these two tests. It is
into account the influence of the intermediate principal interesting to point out that the undrained softening
stress on the stress–strain relationship. response following an anisotropic consolidation is repro-
duced by the model.
3.2.2 Other natural clays
3.3.2 Lower Cromer Till
Undrained triaxial tests in extension were also carried out
on Boston blue clay by Ladd and Varallyay [18] and on Drained and undrained triaxial tests in compression and
Lower Cromer Till by Gens [11]. Figures 9e, f and 10j, k extension with different OCRs on anisotropically consoli-
show the comparisons between measured and predicted dated samples of LCT were conducted by Gens [11].
results of undrained triaxial tests on BBC and LCT. For Simulations were carried out to evaluate the model’s
both natural clays, the present model gives good prediction performance for predicting the compression and extension
for tests in extension (h = 60°) using parameters deter- tests on anisotropically consolidated samples (Fig. 10b, c).
mined from the test in compression (h = 0°). The samples were first anisotropically consolidated under
K0 = 0.5 up to r0a ¼ 350 kPa. Then, they were unloaded
3.3 Influence of consolidation stress ratio along a different stress path to four different over-consol-
idation ratios (OCR = 1, 2, 4, 7) before being sheared in
3.3.1 Boston blue clay compression and extension under undrained condition.
Figure 10b, c shows the comparisons between the experi-
Two undrained triaxial tests on anisotropically consoli- mental data and the model predictions. The comparisons
dated sample of BBC under compression and extension indicate a good agreement in the major features of the
were also performed by Ladd and Varallyay [18]. The undrained behavior for anisotropically consolidated
model parameters determined from tests on isotropically samples.
consolidated samples were used for the predictions.

9
(a) (d) (g) (j)
0.5 400 500 K0=0.4 (Ext)
250 K0=0.5 (Ext)
300 400 K0=0.67 (Ext)
0.45 1 K0=1 (Ext)

q (kPa)
150

q (kPa)

q (kPa)
Model
200 OCR=7 300 K0=0.4 (Com)
e

0.4 OCR=4 K0=0.5 (Com)

100 50 K0=0.67 (Com)
OCR=2 200
IC test K0=1 K0=0.8 (Com)
0.35 OCR=1.5 K0=0.67 K0=1 (Com)
Critical state 1 0 100 K0=0.5 -50
OCR=1
Model K0=0.4
Model Model
0.3 -100 0 -150
10 100 1000 0 100 200 300 400 500 0 5 10 15 -25 -15 -5 5 15
p' (kPa) p' (kPa) 1 (%) (%)
(b) (e) (h) (k) 1
250 400 0
OCR=7(Ext)
OCR=4(Ext) 250
300 1
150 OCR=2(Ext)
q (kPa)

OCR=1(Ext)

q (kPa)
2 150

q (kPa)
q (kPa)

Model 200
OCR=7(Com)
50 OCR=4(Com)
3
50
OCR=2(Com) 100 K0=1
4
-50 OCR=1(Com)
OCR=7 OCR=4 K0=0.67
0 K0=0.5 -50
OCR=2 OCR=1.5 5 K0=0.4
OCR=1 Model Model
-150 -100 6 -150
-25 -15 -5 5 15 0 5 10 15 0 5 10 15 0 50 100 150 200 250
(%) (%)
1 (%) 1
(i)
1
(l) p' (kPa)
(c) (f)
250 0 0.5 0.5
AC line with AC line with
K0 = 1~0.4 K0 = 1~0.4
150 1 0.45
q (kPa)

0.45
(%)

50 2 0.4

e
e
v

0.4
-50 3 OCR=7 OCR=4 0.35
OCR=2 OCR=1.5 Drained tests Undrained tests
OCR=1 Model
-150 4 0.3 0.35
0 100 200 300 400 0 5 10 15 0 100 200 300 400 500 0 50 100 150 200 250
p' (kPa) 1 (%) p' (kPa) p' (kPa)

Fig. 10 Comparison between experimental results and model predictions for tests on LCT clay a void ratio versus mean effective stress for
compression test, b, c undrained tests on anisotropically consolidated samples with different OCRs; d–f drained tests on anisotropically
consolidated samples with different OCRs; g–i drained tests on anisotropically consolidated samples with OCR = 1; j–l undrained tests on
anisotropically consolidated samples with OCR = 1

(a) (b) (c) (d)

2 -5 1.5 -5
Compression Compression Extension Extension
1.5
0 1 0
(%)
(%)

q/p'
q/p'

1
v
v

OCR=1 OCR=2 5 0.5 5

0.5 OCR=1 OCR=2 OCR=1 OCR=2 OCR=1 OCR=2
OCR=4 OCR=8 OCR=4 OCR=8 OCR=4 OCR=8 OCR=4 OCR=8
Model Model Model Model
0 10 0 10
0 5 10 15 0 5 10 15 0 5 10 15 0 5 10 15

d (%) d (%) d (%) d (%)

Fig. 11 Comparison between experimental results and model predictions for drained triaxial tests on Fujinomori clay with different OCRs
a stress ratio q/p0 versus deviatoric strain under compression; b volumetric strain versus deviatoric strain under compression; c stress ratio q/p0
versus deviatoric strain under extension; b volumetric strain versus deviatoric strain under extension

Figure 10d–f shows the comparisons between experi- different values of K0 (0.4, 0.5, 0.67, 1.0) along a different
mental data and model predictions of drained tests on stress path. Then, the samples were subjected to an axial
anisotropically consolidated specimens with different compression until a vertical strain of 15 %. Maximum
OCRs. Similar to the undrained tests, the samples were first shear strength and volume change for samples with dif-
anisotropically consolidated under K0 = 0.5, then unloa- ferent values of OCR are well captured by the present
ded to different values of OCR (1, 1.5, 2, 4, and 7) with model using the same set of parameters as for undrained

10
 (a) (b) (c)
1.5 1.5 1.5

(3J2)0.5/p'
(3J2)0.5/p'
1 1 1
(3J 2)0.5/p'

=0° =0° =0°

=15° =15° =15°
0.5 0.5 0.5
=30° =30° =30°
=45° =45° =45°
Model Model Model
0 0 0
0 5 10 15 -10 -5 0 5 -9 -6 -3 0
1 (%) 2 (%) 3 (%)

Fig. 12 Comparison between experimental results and model predictions for drained true triaxial tests on normally consolidated Fujinomori clay
with different Lode’s angle a stress ratio |q|/p0 versus major principle strain; b stress ratio |q|/p0 versus intermediate principle strain; c stress ratio
|q|/p0 versus minor principle strain

1000 between the numerical and the experimental results using

y= 319.04e -0.047x
the set of parameters given in Table 1.
R² = 0.9728
3.3.3 Softening response under undrained compression
Gp0

100
following anisotropic consolidation

A peculiar behavior worth to be noted is the softening

response in undrained compression for the two cases with
10
0 10 20 30 40 50 K0 consolidation (see K0 = 0.4 and 0.5 in Fig. 10). The
Ip same form of softening response has been observed in
other types of clay (e.g., Fig. 9e, f on Boston blue clay by
Fig. 13 Determination chart for the value of plastic hardening
modulus according to the plasticity index
Ladd and Varallyay [18]). The measured softening
response cannot be attributed to the destructuration process
tests. However, some discrepancies between experiments since the tested clays were reconstituted in the laboratory.
and simulations can be observed in the evolution of the This form of softening response is difficult to model by the
deviatoric stress and volume change with the axial strain. conventional methods using kinematic hardening of the
Four drained triaxial compression tests on normally yield surface (e.g., Ling et al. [21]; Wheeler et al. [33]).
consolidated samples were also selected for simulation. However, using specific rotational kinematic hardening
After being consolidated to different values of K0 (0.4, 0.5, rules and yield-surface shapes, Pestana et al. [25] and
0.67, 1.0), the samples were loaded to failure in drained Dafalias et al. [9] have managed to simulate the softening
condition. Figure 10g–i shows a good agreement between response after K0 consolidation.
the numerical and the experimental results of the drained Different from the approach via kinematic hardening of
triaxial tests, using the set of parameters determined from an the yield surface, the proposed approach employs the
undrained test (Table 1). The measured volumetric strain density state variable ec/e. This assures the void ratio as
increases with the value of K0, which can be explained by the well as the stress state to approach the critical state
position of the compression lines in the e–logp0 plane simultaneously, for any loading path. At large shear strains,
(Fig. 10i). the stress state converges toward the critical state. Thus, the
Undrained triaxial tests on normally consolidated sam- magnitude of the mean effective stress p0 is governed by
ples with different consolidation stress ratios were also the location of the critical state line (in the e–logp0 plane).
simulated. The samples were first anisotropically consoli- The void ratio also approaches the critical state, thus the
dated with four different consolidation stress ratio K0 (i.e.,
 shear strength q (on the p0 –q plane) is determined from p0
the ratio of radial to axial stress r0r r0a ): 0.4, 0.5, 0.67 and corresponding to the critical state ec (see the schematic plot
0.8. Then, for each K0, two subsequent undrained shearing in Fig. 10l). As a consequence, the softening response will
tests were conducted: one in compression (with an increase occur when the deviatoric stress at the end of the K0 con-
in the axial strain) and the other in extension (with a solidation is higher than the undrained shear strength
decrease in the axial strain). For all the eight loading paths determined from the critical state.
mentioned above, Fig. 10j–k shows a good agreement Overall, for all the examples selected in this study, the
numerical simulations are in agreement with the

11
experimental results. For two parameters (Gp0 and D) dif- effective stress, in which the stress ratio increases with
ferent from Cam-Clay models, the value of D varies from the number of cycles; Fig. 14e, f presents the result of a
0.3 to 1.2 and mostly around 1 (see Table 1); and the value constant-amplitude cyclic test under constant mean
of Gp0 is suggested to be determined from the plasticity effective stress. For each test, the curves giving the
index based on widely selected clays (see Fig. 13), thus is stress ratio versus the deviatoric strain and the stress
not needed as input. Therefore, the proposed model ratio versus the volumetric strain are plotted. All com-
incorporating the density state controlling explicitly the parisons between the test results and the numerical
location of the critical state is simple in terms of parame- simulations demonstrate that the model can reasonably
ters determination, and is able to describe drained and well describe the cyclic behavior of clay in drained tri-
undrained behaviors of clay subjected to monotonic load- axial tests.
ing after isotropic or anisotropic consolidation.
3.4.2 Clay mixture
3.4 Cyclic loading
Undrained triaxial tests were conducted on normally con-
3.4.1 Fujinomori clay solidated samples under cyclic loading by Li and Meissner
[20]. The cyclic loading program involves one- and two-
The parameters determined from the drained compression way cyclic tests. All the cyclic tests were stress controlled,
test under monotonic loading (see Table 1; Fig. 11) were with a sinusoidal wave form at frequency of 0.1 Hz. The
used to simulate three drained triaxial tests under cyclic cyclic stress ratio, defined as the ratio of the applied cyclic
loading on Fujinomori clay performed by Nakai and shear stress to the monotonic shearing strength in com-
Hinokio [23]. The model predictions of the clay response pression, ranges from 0.50 to 0.80.
are compared with the experimental results: Fig. 14a, b The set of parameters determined from the monotonic
presents the results of a drained cyclic test under con- tests were used to simulate the undrained cyclic tests. Note
stant confining stress; Fig. 14c, d presents the results of a that the initial slopes of the q  ed curve for monotonic,
varying-amplitude cyclic test under constant mean one- and two-way cyclic tests are different from each

(a) (c) (e)

2 2 2
Constant confining stress Constant p' Constant p'

1 1
1
q/p'
q/p'

q/p'

0 0

0
-1 -1
'3c=196 kPa p'=196 kPa p'=392 kPa
Model Model Model
-1 -2 -2
0 3 6 9 -3 0 3 6 9 0 3 6 9
d (%) d (%) d (%)
(b) (d) (f)
2 2 2
Constant confining stress Constant p' Constant p'

1 1
1
q/p'
q/p'

q/p'

0 0

0
-1 -1
'3c=196 kPa p'=196 kPa p'=392 kPa
Model Model Model
-1 -2 -2
0 3 6 9 0 3 6 9 0 3 6 9
v (%) v (%) v (%)

Fig. 14 Comparison between experimental results and model predictions for drained triaxial tests under cyclic loading on normally consolidated
Fujinomori clay with different stress paths a, b stress ratio q/p0 versus deviatoric and volumetric strains for test at constant confining stress; c,
d stress ratio q/p0 versus deviatoric and volumetric strains for test at constant p0 with increasing q/p0 ; e, f stress ratio q/p0 versus deviatoric and
volumetric strains for test at constant p0 with constant q/p0

12
 (a)
140
Simulation
120

100
q (kPa)

q (kPa)
80

60

40

20

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4

(b) d (% ) d (% )
600
u (kPa)

550

500

Simulation
450
0 5 10 15 20
Number of cycles
(c)
150 Simulation
100
q (kPa)

50
q (kPa)

0
-1.4 -1 -0.6 -0.2 0.2
-50

-100

-150
d (% ) d (% )
(d)
550
u (kPa)

500

450

Simulation
400
0 5 10 15 20
Number of cycles
Fig. 15 Comparison between experimental results and model predictions for undrained tests on clay mixture a deviatoric stress versus deviatoric
strain for one-way cyclic loading; b pore water pressure versus number of cycles for one-way cyclic loading; c deviatoric stress versus deviatoric
strain for two-way cyclic loading; d pore water pressure versus number of cycles for two-way cyclic loading

13
others due to different loading rates. Since the rate- Acknowledgments This research was financially supported by the
dependency behavior of clay was not considered in the opening project of the State Key Laboratory of Geohazard Prevention
and Geoenvironment Protection (Grant No. SKLGP2013K025), the
elastoplastic model, different values of Gp0 were selected National Natural Science Foundation of China (Grant No. 41240024),
for each case (Gp0 = 670 for one-way test, Gp0 = 5,000 the Research Fund for the Doctoral Program of Higher Education of
for two-way test), as done by Li and Meissner [20] and Yu China (Grant No. 20110073120012), and the Shanghai Pujiang Talent
et al. [41]. Simulations are given in Fig. 15a, b for one-way Plan (Grant No. 11PJ1405700).
cyclic tests and in Fig. 15c, d for two-way cyclic tests and
compared with experimental results. The excess pore
References
pressure (Du) and the deviatoric stress are plotted as a
function of the number of cycles and deviatoric strain, 1. Balendran B, Nemat-Nasser S (1993) Double sliding model for
respectively. One can see that the proposed model can cyclic deformation of granular materials including dilatancy
capture with reasonable accuracy, the undrained behavior effects. J Mech Phys Solids 41(3):573–612
of clay subjected to one- and two-way cyclic loading. 2. Been K, Jefferies MG (1985) A state parameter for sands. Geo-
technique 35(2):99–112
3. Biarez J, Hicher PY (1994) Elementary mechanics of soil
behaviour, Balkema
4 Conclusions 4. Bryson LS, Salehian A (2011) Performance of constitutive
models in predicting behavior of remolded clay. Acta Geotech
6:143–154
A simple critical-state-based double-yield-surface model 5. Calladine CR (1971) Microstructural view of the mechanical
was developed for describing the mechanical behavior of properties of saturated clay. Geotechnique 21(4):391–415
clay. The model has two yield surfaces: one for shear 6. Chang CS, Yin Z-Y (2010) Modeling stress-dilatancy for sand
sliding and one for compression. A CSL-related density under compression and extension loading conditions. ASCE J
Eng Mech 136(6):777–786
state was defined to link the peak strength and the phase 7. Chang CS, Yin Z-Y (2010) Micromechanical modelling for
transformation characteristics to the material’s void ratio. inherent anisotropy in granular materials. ASCE J Eng Mech
Therefore, the strain-hardening behavior with contraction 136(7):830–839
for normally consolidated clay and the strain-softening 8. Chowdhury EQ, Nakai T (1998) Consequence of the tij-concept
and a new modelling approach. Comput Geotech 23(4):131–164
behavior with dilation for over-consolidated clay can be 9. Dafalias YF, Manzari MT, Papadimitriou AG (2006) SANI-
modeled. The model guarantees also that stresses and void CLAY: simple anisotropic clay plasticity model. Int J Numer
ratio reach simultaneously the critical state line in the p0 –q– Anal Methods Geomech 30(12):1231–1257
e space. The stress reversal technique was incorporated into 10. Gajo A, Muir Wood D (1999) Severn–Trent sand: a kinematic-
hardening constitutive model: the q–p formulation. Geotechnique
the model for describing the mechanical behavior of clay 49(5):595–614
under loading with changes in stress direction (for instance, 11. Gens A (1982) Stress–strain and strength of a low plasticity clay.
anisotropic consolidation followed by monotonic loading/ Ph.D. Thesis at Imperial College, London University
unloading, cyclic loadings). The model has 6 material 12. Hardin BO (1978) The nature of stress–strain behaviour of soils.
In: Proceedings of the earthquake engineering and soil dynamics,
parameters and 2 state variables which can be easily vol 1, Pasadena, USA, pp 3–90
determined based on one drained or undrained triaxial test 13. Hirayama H (1987) Interpretation of the cam-clay model as a
up to failure with an isotropic consolidation stage. simplified double-yield-surface model. Soils Found 27(3):105–
The capability of the model to reproduce the main fea- 111
14. Hsieh HS, Kavazanjian JE, Borja RI (1990) Double-yield-surface
tures of clay behavior was examined by comparing Cam-clay plasticity model. I: theory. J Geotech Eng 116(9):
experimental results and numerical simulations of drained 1381–1401
and undrained triaxial tests under monotonic loading with 15. Huang WX, Wu W, Sun DA, Scott S (2006) A simple hypoplastic
different conditions (different OCRs, different Lode’s model for normally consolidated clay. Acta Geotech 1:15–27
16. Hujeux JC (1985) Une loi de comportement pour le chargement
angles, different consolidation stress ratios) on various cyclique des sols. In: Davidovici V (ed) Génie Parasismique.
remolded clays (Black clay and White clay) and natural Presses ENPC, France, pp 278–302
clays (Fujinomori clay, Boston blue clay, Lower Cromer 17. Jefferies MG (1993) NorSand: a simple critical state model for
Till, and clay mixture). Drained and undrained tests under sand. Geotechnique 43(1):91–103
18. Ladd CC, Varallyay J (1965) The influence of the stress system
cyclic loadings were also simulated by using the sets of on the behaviour of saturated clays during undrained shear.
parameters determined from monotonic tests. Research Rep. No. R65-11, Department of Civil Engineering,
All comparisons between experimental results and MIT, Cambridge, MA
numerical simulations demonstrate that the proposed 19. Lade PV (2007) Modeling failure in cross-anisotropic frictional
materials. Int J Solids Struct 44(16):5146–5162
model is capable of reproducing the behavior of clays with 20. Li T, Meissner H (2002) Two-surface plasticity model for cyclic
different stress histories, different stress paths, different undrained behavior of clays. ASCE J Geotech Geoenviron Eng
drainage conditions and different loading conditions. 128(7):613–626

14
21. Ling HI, Yue D, Kaliakin VN (2002) Anisotropic elastoplastic 32. Vermeer PA (1978) A double hardening model for sand. Geo-
bounding surface model for cohesive soils. ASCE J Eng Mech technique 28(4):414–433
128(7):748–758 33. Wheeler SJ, Näätänen A, Karstunen M, Lojander M (2003) An
22. Mašin D (2005) A hypoplastic constitutive model for clays. Int J anisotropic elasto-plastic model for soft clays. Can Geotech J
Numer Anal Methods Geomech 29(4):311–336 40:403–418
23. Nakai T, Hinokio M (2004) A simple elastoplastic model for 34. Whittle AJ, Kavvadas MJ (1994) Formulation of MIT-E3 con-
normally and overconsolidated soils with unified material stitutive model for overconsolidated clays. J Geotech Eng
parameters. Soils Found 44(2):53–70 120(1):173–198
24. Ohmaki S (1979) A mechanical model for the stress–strain 35. Yamakawa Y, Hashiguchi K, Ikeda K (2010) Implicit stress-
behaviour of normally consolidated cohesive soil. Soils Found update algorithm for isotropic Cam-clay model based on the
19(3):29–44 subloading surface concept at finite strains. Int J Plast
25. Pestana JM, Whittle AJ, Gens A (2002) Evaluation of a consti- 26(5):634–658
tutive model for clays and sands: Part II—Clay behaviour. Int J 36. Yao YP, Hou W, Zhou AN (2009) UH model: three-dimensional
Numer Anal Methods Geomech 26(11):1123–1146 unified hardening model for overconsolidated clays. Geotech-
26. Roscoe KH, Burland JB (1968) On the generalized stress–strain nique 59(5):451–469
behavior of ‘wet’ clay. Engineering Plasticity. Cambridge Uni- 37. Yin Z-Y, Chang CS (2009) Microstructural modelling of stress-
versity Press, Cambridge, pp 553–609 dependent behaviour of clay. Int J Solids Struct 46(6):1373–1388
27. Schofield AN, Wroth CP (1968) Critical state soil mechanics. 38. Yin Z-Y, Chang CS (2013) Stress-dilatancy for sand under
Mcgraw Hill, Maidenhead loading and unloading conditions. Int J Num Anal Methods
28. Schweiger H, Wiltafsky C, Scharinger F, Galavi V (2009) A Geomech. doi:10.1002/nag.1125
multilaminate framework for modelling induced and inherent 39. Yin Z-Y, Chang CS, Hicher PY, Karstunen M (2009) Micro-
anisotropy of soils. Geotechnique 59(2):87–101 mechanical analysis of kinematic hardening in natural clay. Int J
29. Sheng D, Sloan SW, Yu HS (2000) Aspects of finite element Plast 25(8):1413–1435
implementation of critical state models. Comput Mech 26:185– 40. Yin Z-Y, Chang CS, Hicher PY (2010) Micromechanical mod-
196 elling for effect of inherent anisotropy on cyclic behaviour of
30. Stallebrass SE, Taylor RN (1997) The development and evalua- sand. Int J Solids Struct 47(14–15):1933–1951
tion of a constitutive model for the prediction of ground move- 41. Yu H-S, Khong C, Wang J (2007) A unified plasticity model for
ments in overconsolidated clay. Geotechnique 47(2):235–253 cyclic behaviour of clay and sand. Mech Res Commun 34:97–114
31. Taiebat M, Dafalias YF (2008) SANISAND: simple anisotropic 42. Zervoyannis C (1982) Etude synthetique des proprietes meca-
sand plasticity model. Int J Numer Anal Methods Geomech niques des argiles et des sables sur chemins oedometrique et
32(8):915–948 triaxial de revolution. Ph.D. thesis, Ecole Centrale de Paris

15